Tested Gemma-4 E2B-it on Raspberry Pi 4.
the way to convert Gemma-4 E2B-it to gguf
models
https://huggingface.co/baxin/gemma-4-E4B-it-E2B-it-Q4_K_M
llama.cpp
Run Gemma-4 E2B-it with llama.cpp on Raspberry Pi4
Tested Gemma-4 E2B-it on Raspberry Pi 4. the way to convert Gemma-4 E2B-it to gguf Quantizing...
TL;DRAI
Gemma-4 E2B-it (Q4_K_M) tested on Raspberry Pi 4 with llama.cpp reaches ~1.8 t/s generation — too slow for agentic workloads; LFM2.5-8B drops to just 0.5 t/s. At ~$305, a Pi 5 loses on price/performance vs. a $300–400 mini PC with 16GB RAM for sub-10B local inference.
774 words~4 min read
